PaperMind - Multimodal Grounding: leaderboard

PaperMind (2026) tests reasoning over real scientific papers from seven domains, with answers rated 1 to 5 by a GPT-4o judge against references taken from the papers. Here the model writes the caption of a figure given the introduction of the paper, compared with the original caption; higher is better.

Metric: Judge score for writing the caption of a figure from a real paper given the paper's introduction, the original caption as reference; rated by a GPT-4o judge on a 5-point scale (1-5) against the reference, averaged over the seven scientific domains; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 7 models tracked. Updated 14 Nov 2026.

Top models

#ModelScoreOverall rank
1Gemini 2.5 Pro2.39#276
2GPT-4o Mini2.18#1667
3Claude 3.5 Sonnet2.14#752
4Qwen 3 VL 4B Instruct2.11#1332
5Gemma 3 4B (IT)1.89#4139
6Claude 3 Haiku1.87#2663
7Phi-3.5-Vision Instruct1.75#6268

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=papermind-multimodal-grounding · How It Works · Data refreshed daily, snapshot 2026-11-21.